Statistical Modelling 4 (2004), 113126
Estimating intraclass correlation for binary data
using extended quasi-likelihood
Youngjo Lee
Department of Statistics,
Seoul National University,
Seoul,
South Korea.
Abstract:
Extended quasi-likelihood has been recommended for the estimation
of dispersion parameters when there is no likelihood available.
It was originally developed for models with independent errors and
can have poor efficiency in models with correlated errors. We show
how the generalization of extended
quasi-likelihood to such models can maintain high efficiency.
Keywords:
beta binomial; extended quasi-likelihood; hierarchical likelihood; marginal likelihood;
pseudo-likelihood
Downloads:
Data and
Genstat code in zipped archive
back